With complex supply chains that stretch overseas, even companies that vow to keep a close eye on their main suppliers often fail to eradicate human-rights violations at all levels of production. Worse, these abuses (child labor, abominable working conditions, and wage theft to name a few) are often ignored until a tragedy occurs. Afterward, there are always pledges and plans to reform in order to ensure that such atrocities never happen again.

But in the wake of such tragedy, such as Rana Plaza, does substantive change soon follow? A new report from NYU’s Stern Center for Business and Human Rights suggests it doesn’t.

For the report, Sarah Labowitz, a co-director of the center, and Dorothée Baumann-Pauly, the director of research, parsed public data about factories and labor practices in Bangladesh. They also conducted on-the-ground surveys in Dhaka, the site of the 2014 Tazreen Factory fire and near the site of the 2013 Rana Plaza collapse, which killed more than 1,100 workers when a garment factory building collapsed. In that incident, unsafe and illegal working conditions had made it impossible for many to safely evacuate when the building became unstable and dangerous.

What researchers found more than two years after the Rana Plaza disaster remains disturbing and disheartening: Of the 3,425 inspections that took place in Bangladesh, only eight factories have remedied violations enough to pass a final inspection. And that’s despite commitments of more than $280 million from the international community to clean up the industry in Bangladesh.

The NYU researchers have some thoughts on why change remains so slow and incomplete. Part of the problem is sheer size of the industry, along with the lack of reliable data. Researchers found a total of 7,000 factories linked to export markets in Bangladesh. Previous estimates had put that figure much lower, at around 4,500. The discrepancies are largely because the depth of subcontracting was profound, with many registered factories relying on labor and goods from informal factories who don’t show up on official production rosters. These make up about half of all factories in the country, and one-third of the factories in Dhaka.

That’s particularly troubling because these indirect suppliers are harder to assess and control and are consistently involved in both major supply chains and garment factory tragedies. They aren’t registered with government entities or officially listed with brands, creating class of workers who are nearly invisible--and thus hard to protect via regulations and audits. Additionally, these shadow suppliers are often smaller and rely on more intense human labor so they can offer lower prices, undercutting official factories that are regulated and bearing the cost of paying workers minimum wage while keeping factories compliant with international and local standards. Further, it’s sometimes hard for suppliers to know whether their factories are subcontracting to these shadow suppliers.

In total, NYU’s report found about 5.1 million factory workers in Bangladesh, about 1 million more than estimated by the International Labour Organization, the organization responsible for much of the oversight and data gathering about global supply chains. In light of these adjusted figures, it’s clear that much of the progress touted by Bangladesh’s government—such as the 80 percent factory safety approval rate—is overstated, since around half of the factories involved in the creation of exports aren’t even on official rosters.

“The problems presented by an under-regulated and informal garment sector are much bigger than anyone has acknowledged up to this point,” the report’s authors write. But unfortunately,the vastness (and undercounting) of the supply chain isn’t the only problem.

In the wake of Rana Plaza, factory safety programs meant to improve fire and building protocols were agreed upon by politicians and several major brands such as Costco, Walmart, and Macy’s. But according to the NYU report, these agreements only cover 27 percent of factories and 45 percent of laborers who work for some of the largest and most advanced factories in the country. And for the most part, even the factories that fall under the purview of these programs haven’t completed the reforms necessary to make them safe. The holdup, the study finds, is a battle over who should pay for expensive electrical and structural upgrades—the factories or the brands that buy from them.

The failure to reform isn’t just bad (and dangerous) for workers. Both reformers and brands are discouraged by the delayed progress, and when the agreements in place to better Bangladesh’s factories expire in 2018, companies may simply exit the market in favor of less heavily scrutinized production centers And the loss of the garment sector, which has been responsible for so much of Bangladesh’s economic progress, could be catastrophic for the country’s economy.

Though the report shows that there is a very long way to go, the possibility of creating a safer garment production system isn’t impossible. It will just require significant investment—of time, manpower, and money—by brands, factories, and the governments that benefit from the industry. Getting all of those groups on the same page, may prove to be the most difficult (and political) task of all. But change must happen, because in the meantime, workers continue to toil in inhumane conditions.